Third Circuit upholds Thomson Reuters' win over Ross Intelligence: first US appellate ruling rejecting fair use for AI training
On Sept 29, 2026 the US Court of Appeals for the Third Circuit affirmed a February 2025 ruling that Ross Intelligence infringed Thomson Reuters' Westlaw headnotes by using material derived from them to build an AI legal-search tool, and that this was not fair use. It is the first US appellate decision on whether training AI on copyrighted material can be fair use. The opinion was filed under seal, with redactions due within 10 days, so the reasoning is not yet public.
Key facts
- Court: US Court of Appeals for the Third Circuit (Philadelphia), interlocutory appeal from the District of Delaware; decided Sept 29, 2026
- Affirmed: Judge Stephanos Bibas's February 2025 summary judgment that Ross copied Westlaw headnotes and was not entitled to a fair-use defense, because it used them for the same purpose as Westlaw and to compete with it
- Opinion filed temporarily under seal; parties must propose redactions with reasons within 10 days (LawSites). Chat GPT Is Eating the World reports Judge Montgomery-Reeves wrote the panel opinion
- The case dates to 2020; Ross Intelligence has since shut down
- Reuters: a first-of-its-kind appellate ruling in the wave of AI-training copyright cases; MLex: 'First US appellate finding on fair use for AI training favors copyright holders'
- Caveat: Ross built a non-generative search tool that competed directly with Westlaw; district courts in 2025 (Bartz v. Anthropic, Kadrey v. Meta) found generative-model training fair use on different facts
What happened
Thomson Reuters sued Ross Intelligence in 2020, alleging that Ross trained its legal research engine on Westlaw headnotes (summaries of points of law) obtained through a third party. Judge Bibas (sitting by designation) ruled for Thomson Reuters in February 2025 and certified the questions for interlocutory appeal. The Third Circuit affirmed on Sept 29, 2026.
Why it matters
Dozens of AI copyright suits (authors, news publishers, music labels) turn on fair use. This is the first appellate ruling, and it went against the AI developer. Its reach for generative models depends on the sealed reasoning. Check the unsealed opinion (expected in October 2026) for whether the court limits its holding to competing, non-generative uses.
Changelog
- 2026-09-30: created (sweep 2026-09-30)
Sources (6)
- pressReuters: US appeals court upholds Thomson Reuters' landmark win in AI training lawsuit
- pressLaw360: 3rd Circ. affirms Thomson Reuters' Westlaw AI copyright win
- pressLawSites: 3rd Circuit issues opinion in Thomson Reuters v. ROSS, but for now it is sealed
- discussionChat GPT Is Eating the World: Third Circuit affirms rejection of Ross's fair use defense
- pressMLex: First US appellate finding on fair use for AI training favors copyright holders
- pressMediaPost: Appeals court sides with Thomson Reuters in battle over AI training
id: 2026-09-29-third-circuit-thomson-reuters-ross-fair-use · updated 2026-09-30 · open in the interactive timeline